PannotatoR Examples

Anonymised for review

Data analysis for PannotatoR paper

Install libraries

library(dplyr)
library(ggplot2)
library(mapview)
library(RColorBrewer)
library(rcompanion)
library(sf)
library(stringr)
library(webshot)

Upload species richness, spinifex cover and tree health .rds files

# Read in the .Rds files containing species, spinifex cover and tree crown data

# If you have exported an .Rds file from PannotatoR
df_species <- readRDS("Data_files/df_species.Rds")

# If you have exported an .csv file from PannotatoR
df_spinifex <- read.csv(file = "Data_files/df_spinifex.csv", header = TRUE, sep = ",")

df_oak <- readRDS("Data_files/oak_data_user1.Rds")

Add plot and frame numbers to each .rds file

# Now we add the plot and information to each frame in the .rds file. This is useful code if you have multiple plots or subplots in a single .rds file. In this example we have only one plot (plot_06_new). A .csv file linking plot ID to frame or .kmz file ID is required

# Import the csv file 
df_plot_number <- read.csv(file = "Data_files/Plot_names.csv") 

# This function adds a plot_kmz column,fills it with a substring from the imagefile column, add a frame column with a substring from the imagefile column, and then match the plot number to the plot_kmz in the df_plot_number dataframe.

add_plot_kmz <- function(df) {
  df$plot_kmz <- substr(df$imagefile,1,24)
  df$frame <- substring(df$imagefile,1,37 )
  df$plot_name_new <- df_plot_number$plot_name_new[match(df$plot_kmz, df_plot_number$plot)]
  return(df)
}

df_species <- add_plot_kmz(df_species)
df_spinifex <- add_plot_kmz(df_spinifex)
df_oak <- add_plot_kmz(df_oak)

Map the location of data points in each file

# Using the geometry field
df_species <- st_as_sf(df_species, wkt = "geometry",crs = 4326) 
mapview(df_species)

df_spinifex <- st_as_sf(df_spinifex, wkt = "geometry",crs = 4326) 
# mapview(df_spinifex)

df_oak <- st_as_sf(df_oak, wkt = "geometry",crs = 4326) 
# mapview(df_oak)

Analysis of species distribution and richness data

Clean up the species data (df_species) for mapping and analysis

# View(df_species)
# Create a new species variable 
df_species$Species_new <- NA

# If variable dd3 = -999, then add species (mulga information), otherwise keep the species listed in dd3. This annoatation was used to quantify mulga cover and also to add additional species #
df_species$Species_new <- ifelse(df_species$dd3 == -999, df_species$species, df_species$dd3)


# For simplicity of data analysis we must now rename the species in the species_new variable 

df_species$Species_new[df_species$Species_new == 'Acacia_aneura (mulga)'] <-'Acacia_aneura'

df_species$Species_new[df_species$Species_new == 'Acacia_ligulata (dune wattle)'] <-'Acacia_ligulata'

df_species$Species_new[df_species$Species_new == 'Acacia_minyura (desert mulga)'] <-'Acacia_minyura'

df_species$Species_new[df_species$Species_new == 'Allocasuarina_decaisneana (oak)'] <-'Allocasuarina_decaisneana'

df_species$Species_new[df_species$Species_new == 'Burnt Mulga Mature (3-6m)'] <-'Acacia_aneura'

df_species$Species_new[df_species$Species_new == 'Burnt Mulga Juvenile (1-3m)'] <-'Acacia_aneura'

df_species$Species_new[df_species$Species_new == 'Burnt Mulga Overmature (3-6m)'] <-'Acacia_aneura'

df_species$Species_new[df_species$Species_new == 'Codonocarpus_cotinifolius (poplar)'] <-'Codonocarpus_cotinifolius'

df_species$Species_new[df_species$Species_new == 'Grevillea_eriostachya (honey grevillea)'] <-'Grevillea_eriostachya'

df_species$Species_new[df_species$Species_new == 'Grevillea_stenobotrys (sandhill grevillea)'] <-'Grevillea_stenobotrys'

df_species$Species_new[df_species$Species_new == 'Gyrostemon_ramulosus (camel poison)'] <-'Gyrostemon_ramulosus'

df_species$Species_new[df_species$Species_new == 'Mulga Juvenile (1-3m)'] <-'Acacia_aneura'

df_species$Species_new[df_species$Species_new == 'Mulga Mature (3-6m)'] <-'Acacia_aneura'

df_species$Species_new[df_species$Species_new == 'Mulga Seedling (<1m)'] <-'Acacia_aneura'

df_species$Species_new[df_species$Species_new == 'Santalum_lanceolatum (sandalwood)'] <-'Santalum_lanceolatum'

df_species$Species_new[df_species$Species_new == 'Themeda_australis (kangaroo grass)'] <-'Crotalaria_cunninghamii'

df_species$Species_new[df_species$Species_new == 'Themeda_avenaceae (oat grass)'] <-'Crotalaria_cunninghamii'

df_species$Species_new[df_species$Species_new == 'Acacia_kempeana (granite wattle)'] <-'Acacia_kempeana'

df_species$Species_new[df_species$Species_new == 'Corymbia_opaca (desert bloodwood)'] <-'Corymbia_opaca'

df_species$Species_new[df_species$Species_new == 'Eucalyptus_gamophylla (blue mallee)'] <-'Eucalyptus_gamophylla'

# We now Clean up a few more cases. We don't want any data points that contain no species data, unknown species, or species with unclear taxonomy (e.g., Hakea spp.) 

df_species <- subset(df_species, Species_new != "No Mulga present")
df_species <- subset(df_species, Species_new != "Re_Check_Unknown_Shrub")
df_species <- subset(df_species, Species_new != "Hakea_sp")
df_species <- subset(df_species, Species_new != "Acacia_sp")
df_species <- subset(df_species, Species_new != "Gyrostemon_sp.")

# Remove species and location duplicates 
duplicated_rows <- duplicated(df_species[c("Species_new", "geometry")])
df_species<- df_species[!duplicated_rows, ]

# We must now jitter the geometries so that we can see the data easier in mapview. Many of the locations are very close together 
df_species_jittered <- st_jitter(df_species, amount = 0.00001)

Map the species distributions (generate image for Figure 4a in paper)

This is useful code for plotting species distributions in mapview, including setting colour schemes within a user-specified species order.

# This code generates Figure 4A 

# Confirm the data 
#mapview(df_species_jittered, zcol = "Species_new", cex = 8, alpha = 0.9) 

# Here I reorder the species from sand dune crest species to swale species (approximately) 

df_species_jittered$speciesReorder <- factor(df_species_jittered$Species_new, c('Crotalaria_cunninghamii','Gyrostemon_ramulosus', 'Acacia_ligulata',   'Aluta_maisonneuvei', 'Newcastelia_spodiotricha', 'Gyrostemon_tepperi','Codonocarpus_cotinifolius', 'Grevillea_eriostachya', 'Allocasuarina_decaisneana', 'Exocarpus_sparteus',  'Eremophila_sp', 'Eucalyptus_Mallee_sp.', 'Eucalyptus_gamophylla', 'Corymbia_opaca', 'Acacia_kempeana' ,'Acacia_aneura'))

# Here I colour species groups according to their typical ecological position; e.g., sand dune crest species in red, swale species in white
p <- mapview(df_species_jittered, zcol = "speciesReorder", cex = 5, alpha = 1, alpha.regions = 0.9, layer.name = "Species", col.regions = c("red4","red4", 'red4', "orange", 'yellow3','yellow3', 'yellow2','yellow2','green3','green2','green4', 'deepskyblue2', 'deepskyblue2','purple','white','white'), 
        na.rm = TRUE)

mapshot(p, file = "Data_files/Fig_4a.png")
p

Make a 25 m buffer around each plot, create a polygon, segment it and calculate species diversity in each segment

This is useful code to add a buffer around each data point in the dataframe, combined these together if overlapping into a series of polygons. The intent is to then split up the polygons according to another spatial dataframe, in this case a .kml containing lines that divide the west side of the transect into 50 m segments with approximately homogeneous vegetation.

Make buffer around each point and create polygon(s)

# Add a spatial buffer of 25m around each point
buffer <- st_buffer(df_species_jittered, dist = 50)

# Combine the resulting buffers into a single polygon
polygon <- st_union(buffer)

# mapview(polygon) # we can see that the buffers have been merged into a series of polygons.

# Export the resulting polygon as a .kml file
st_write(polygon, "Data_files/buffer_plot_polygon.kml", append= FALSE, driver = "KML")

Import the .kml file for segmenting the polygon(s) and cut the polygon(s)

# This is.kml file that contains lines that cut the westerly side of the study transect into a series of segments roughly 50 m apart. Each segment was oriented to contain homogenous vegetation. It was drawn in Google Earth. 

segmentation_lines <- st_read("Data_files/polygon_cuts.kml")

# Map the lines 
# mapview(segmentation_lines) 

# Before we cut the polygons using the segmentation lines we mustrepair the geometry of the polygon and lines data for any errors 
polygon_valid <- st_make_valid(polygon)
# mapview(polygon_valid)

lines_valid <- st_make_valid(segmentation_lines)
# mapview(lines_valid)

# Now cut the polygon using the segmentation lines 
cut_polygon <- polygon_valid %>% 
  lwgeom::st_split(lines_valid) %>% 
  st_collection_extract("POLYGON")

mapview(cut_polygon)

Calculate the species richness in each of the new polygons

# Now we will calculate the number of species in each polygon segment.We are interested in polygons with IDs.

# The cut_polygon object is an 'sfc' object and we want to make it an 'sf' object
class(cut_polygon) 

# Create a new data frame
df <- data.frame(id = 1:length(cut_polygon)) 

# Add the sfc_POLYGON object as a geometry column
df$geometry <- cut_polygon 

# Convert the data frame to an sf object
cut_polygon_sf <- st_as_sf(df)  

# Now an sf object with 72 polygons.
# str(cut_polygon_sf)

st_write(cut_polygon_sf, "Data_files/cut_polygon_plot_sf.kml", append= FALSE, driver = "KML")

# Now determine which polygon each point falls into 
result <- st_join(df_species_jittered, cut_polygon_sf)

# Convert the polygon ID to character 
result$id <- as.character(result$id) 

# Now we calculate the species diversity of each polygon 

species_div <- result %>%
  group_by(id) %>%
  summarise(num_species = n_distinct(speciesReorder))

species_div

Plot the centroids of each polygon with species richness of the polygon (image for Figure 4f)

# To enable plotting we calculate the centroids of each polygon
average_geometry <- st_centroid(species_div$geometry)
species_div_centroids <- st_set_geometry(species_div, average_geometry)

# mapview(species_div_centroids)

# We can select the centroids along the west side of the transect using this code if we want, default is to plot all

# id_values <- c(14:23, 25:38, 2:7, 9)
# Species_div_centroids_subset <- species_div_centroids[species_div_centroids$id %in% id_values, ]

# Rename
species_div_centroids_subset <- species_div_centroids 

# Removed the very large unsegmented polygon
species_div_centroids_subset <- species_div_centroids_subset[species_div_centroids_subset$num_species < 7, ]

# Check the data
# mapview(species_div_centroids_subset)

mapviewOptions(basemaps = c("Esri.WorldImagery"),
               raster.palette = grey.colors,
               vector.palette = colorRampPalette(c("white", "red3")),
               na.color = "gray",
               layers.control.pos = "topright")

q <- mapview(species_div_centroids_subset, zcol = "num_species", layer.name = "Richness", cex = 8, alpha = 0.9, alpha.regions = 0.9, na.rm = TRUE)

mapshot(q, file = "Data_files/Fig_4f.png")

q

Analysis of spinifex data

Check the data and correct outliers

Here we are interested in analysing two variables: the % cover of live and dead material within the 10 m diameter plots.

# Ensure variables are numeric
df_spinifex$live_cover<- as.numeric(df_spinifex$live_cover)
df_spinifex$dead_cover<- as.numeric(df_spinifex$dead_cover)

# Check the range of the data fields of interest. Percentage cover is the % cover within the 10 m diameter plots, impact score is based on the percentage of dead versus live material.

# Within range
summary(df_spinifex$live_cover) 
# Within range
summary(df_spinifex$dead_cover) 

# Generate an object with total live and dead cover in each frame 
frame_sum <- df_spinifex %>% 
  group_by(frame) %>%
  summarise(across(c(dead_cover, live_cover), sum, .names = "{.col}"))

# Calculate the total spinifex cover variable 
frame_sum$total_cover <- frame_sum$dead_cover+ frame_sum$live_cover 

frame_sum

Map the total cover of live and dead cover within each image frame (images for Figures 5c and 5d)

# Rename the object
data <- frame_sum  

# Set the mapview options for live cover
mapviewOptions(basemaps = c("Esri.WorldImagery"),
               raster.palette = grey.colors,
               vector.palette = colorRampPalette(c("transparent", "green", 'darkgreen')),
               na.color = "gray",
               layers.control.pos = "topright")

# Plot of live cover 
p <- mapview(data, zcol = "live_cover", layer.name = "Cover(%)", cex = 10, alpha = 0.9, alpha.regions = 0.9,  na.rm = TRUE)

p

mapshot(p, file = "Data_files/Fig_5c.png")

# Set the mapview options for dead cover
mapviewOptions(basemaps = c("Esri.WorldImagery"),
               raster.palette = grey.colors,
               vector.palette = colorRampPalette(c("white", "red", "darkred")),
               na.color = "gray",
               layers.control.pos = "topright")

# Plot of dead cover
q <- mapview(data, zcol = "dead_cover", layer.name = "Cover(%)", cex = 10, alpha = 0.9, alpha.regions = 0.9, na.rm = TRUE)

mapshot(q, file = "Data_files/Fig_5d.png")

q

Map the cover of different spinifex size classes within each image frame (image for Figure 5e)

Re-name spinifex species/size data for size class analysis

df_spinifex$species[df_spinifex$species == 'No Triodia Present'] <- 'Absent'

df_spinifex$species[df_spinifex$species == 'Triodia pungens  >100cm old fragmentary'] <- 'frag'

df_spinifex$species[df_spinifex$species == 'Triodia sp. >100cm old fragmentary'] <- 'frag'

df_spinifex$species[df_spinifex$species == 'Triodia pungens < 50 cm'] <- '<50'

df_spinifex$species[df_spinifex$species == 'Triodia sp. 50-<100 cm'] <- '50_100'

df_spinifex$species[df_spinifex$species == 'Triodia sp. 100-<200 cm'] <- '100_200'

df_spinifex$species[df_spinifex$species == 'Triodia pungens 200-<400 cm'] <- '200_400'

df_spinifex$species[df_spinifex$species == 'Triodia sp. < 50 cm'] <- '<50'

df_spinifex$species[df_spinifex$species == 'Triodia pungens 50-<100 cm'] <- '50_100'

df_spinifex$species[df_spinifex$species == 'Triodia pungens 100-<200 cm'] <- '100_200'

df_spinifex$species[df_spinifex$species == 'Triodia schinzii 50-<100 cm'] <- '50_100'

df_spinifex$species[df_spinifex$species == 'Triodia pungens 400+ cm'] <- '>400'

Calculate and map the cover and distribution of < 50cm and fragmentary (old > 1m) size classes (image for Figure 5e)

# Subset spinifex data for only < 50 cm size class
Less50cm <- subset(df_spinifex, df_spinifex$species == "<50")
 
# Subset spinifex data for only old fragmentary size class
frag <- subset(df_spinifex, df_spinifex$species == "frag")

# Jitter the data so that they can be seen more easily in the map
jittered_frag <- st_jitter(frag, amount = 0.0001)  
jittered_Less50cm <- st_jitter(Less50cm, amount = 0.0001)

# Map the variables together - Figure 5D 
p <- mapview(jittered_frag, zcol = "percentage_cover", layer.name = "Frag_Cover(%)", cex = 8, alpha = 0.9, alpha.regions = 0.75, col.regions = colorRampPalette(c("yellow", "orange", 'orange4')), na.rm = TRUE)  + mapview(jittered_Less50cm, zcol = "percentage_cover", cex = 8, alpha = 0.75, alpha.regions = 0.9, col.regions = colorRampPalette(c("lightblue", "blue", 'darkblue')),layer.name = "L50_Cover(%)", na.rm = TRUE)

mapshot(p, file = "Data_files/Fig_5e.png")

p

Create boxplot of size class vs pre-drought and post-drought percentage cover with kruskal-wallis test (image for Figure 5f)

df_spinifex_full_data <- df_spinifex

df_spinifex <- df_spinifex %>%
  filter(percentage_cover > 0)

df_spinifex <- df_spinifex[df_spinifex$species != 'Absent', ]
df_spinifex <- df_spinifex[df_spinifex$species != 'NA', ]

df_spinifex$speciesReorder <- factor(df_spinifex$species, c("<50", "50_100", "100_200", "200_400", ">400", "frag"))

p <- ggplot(df_spinifex, aes(x = speciesReorder, y = percentage_cover)) +
  geom_boxplot() + theme_bw() +  theme(panel.grid.major = element_blank(),
        panel.grid.minor = element_blank()) +
  theme(axis.text.x = element_text(size = 14),
        axis.text.y = element_text(size = 14))

p <- p + ggtitle("") +
  xlab("Size class (cm)") + ylab("Pre-drought cover") +
  theme(axis.text=element_text(size=12), axis.title=element_text(size=14,face="bold"))

ggsave("Data_files/Fig_5f.png", width = 6.81, height = 6.45, dpi = 300)

# Test for nonparametric relationship between percentage cover and size class

kruskal.test(percentage_cover ~ speciesReorder, data = df_spinifex)

Analysis of desert oak crown health data

Import the lookup csv containing size classes and match to size class data in the df_oak object

# The original df_oak data have been imported in code above 

# Import the match table
df_oak_class <- read.csv("Data_files/Oak_sizes_reclass.csv")

# We must create new variables to allow the match 
df_oak$Taxon <- NA
df_oak$Size_class <- NA
df_oak$Burnt <- NA
df_oak$Count <- NA

# We must remove whitespace and other incompatibilities

df_oak$species <- str_replace_all(df_oak$species, "<.*?>", "")
df_oak_class$species <- str_replace_all(df_oak_class$species, "<.*?>", "")

df_oak$species <- str_to_lower(df_oak$species)
df_oak$species <- str_replace_all(df_oak$species, "\\s+", "")

df_oak_class$species <- str_to_lower(df_oak_class$species)
df_oak_class$species <- str_replace_all(df_oak_class$species, "\\s+", "")

# Now merge Taxon, Burnt, Size_class, Species

df_oak$Taxon <- df_oak_class$Taxon[match(df_oak$species, df_oak_class$species)]

df_oak$Burnt <- df_oak_class$Burnt[match(df_oak$species, df_oak_class$species)]

df_oak$Count <- df_oak_class$Count[match(df_oak$species, df_oak_class$species)]

df_oak$Size_class <- df_oak_class$Size_class[match(df_oak$species, df_oak_class$species)]

Calculate crown health score and compare crown health across size classes (create image for Figure 6f)

# Ensure that the variables are numeric 
df_oak$dd1 <- as.numeric(df_oak$dd1) 
df_oak$dd2 <- as.numeric(df_oak$dd2)
df_oak$dd3 <- as.numeric(df_oak$dd3)

# Calculate the crown score (HC = AC × CE × CD in paper)
df_oak$Crown_score <- df_oak$dd1/100 * df_oak$dd2/100 * df_oak$dd3/100

# Reorder the classes in order of size 
df_oak$Size_class_Reorder <- factor(df_oak$Size_class, c("Seedling", "Juvenile", "Early_Pole", "Late_Pole", "Early_Mature", "Mature"))

# Boxplot of size class and burnt vs crown score
p <- ggplot(df_oak, aes(x = Size_class_Reorder, y = Crown_score,  z = Burnt, fill = Burnt, color = Burnt )) +
  geom_boxplot() + theme_bw() + theme(panel.grid.major = element_blank(),
        panel.grid.minor = element_blank()) +
      scale_fill_manual(values=c("green4","brown3")) +
      scale_color_manual(values=c("darkgreen","brown4")) +
      theme_classic()  +
      theme(panel.border = element_rect(color = "black", fill = NA, size = 2))

p <- p + ggtitle("") +
  xlab("Size class") + ylab("Crown score") +
  theme(axis.text=element_text(size=12), axis.title=element_text(size=14,face="bold"))

ggsave("Data_files/Fig_6f.png", width = 7.25, height = 6.45, dpi = 300)

# Test for nonparametric relationship between crown score and burnt + size class 
scheirerRayHare(Crown_score ~ Burnt + Size_class_Reorder, data = df_oak)

Map burnt and unburnt tree crowns (image for Figure 6e)

# Map burnt and unburnt areas 
m <- mapview(df_oak, zcol = "Burnt", cex = 8, alpha = 0.9, alpha.regions = 0.9, layer.name = "Burnt", col.regions = c("green2", "red2"), na.rm = TRUE)

mapshot(m, file = "Data_files/Fig_6e.png")

m

Map the distribution of tree size classes (image for Figure 6c)

# Map size classes 
p <- mapview(df_oak, zcol = "Size_class_Reorder", cex = 8, alpha = 0.9, alpha.regions = 0.9, layer.name = "Size_class", col.regions = c("#389f0a", "#3F704D","yellow", "yellow3", "red2", "red4"),na.rm = TRUE)

mapshot(p, file = "Data_files/Fig_6c.png")
p

Plot the crown health of mature and early mature size classes (Figure 6d)

# Select only mature and early mature size classes 

Mature <- subset(df_oak, df_oak$Size_class == "Mature" | df_oak$Size_class == "Early_mature")

mapviewOptions(basemaps = c("Esri.WorldImagery"),
               raster.palette = grey.colors,
               vector.palette = colorRampPalette(c("red", "yellow", "green2")),
               na.color = "gray",
               layers.control.pos = "topright")

q <- mapview(df_oak, zcol = "Crown_score", layer.name = "Crown_score", cex = 8, alpha = 0.9, alpha.regions = 0.9, na.rm = TRUE)

mapshot(q, file = "Data_files/Fig_6d.png")
q